Topic Affinity Analysis for an Astronomy and Astrophysics Data Set

In this paper we map the affinity between topics extracted from a body of literature published in Astronomy and Astrophysics journals between 2003-2010. The topics are extracted using the popular information theoretic Infomap clustering algorithm (Rosvall & Bergstrom, 2008) iteratively on the giant component of the direct citation network constructed from the data. The affinity network shows what topics are disproportionally well connected (by citations) to other topics. The topology of the network highlights a large division into astrophysics versus astronomically oriented publications. Bridging between those two domains is a population of smaller topics. Going forward, we plan to create and analyze topic affinity network maps for alternative solutions to the topic extraction challenge on that same data set that are produced by our colleagues and that will be discussed and compared at the proposed special session on 'Same data? Different results? The performative nature of algorithms for topic detection in science' at ISSI 2015. We expect that topic affinity mappings will help to examine the nature of differences between different topic extraction solutions. Conference Topic Methods and techniques (special session on algorithms for topic detection) Introduction The mapping of research topics and collaborative ties in scientific research fields (Morris 2008) is flourishing for a number of reasons. Increasingly, scholarly publications and their metadata are available from a variety of sources (digital libraries, institutional and disciplinary repositories, along with bibliographic abstracting services such as the long established Web of Knowledge and more recently, Scopus). Complementing this is the emergence of sophisticated algorithms for the analysis of complex networks (Newman 2003b) and the wide availability of advanced user-friendly network analysis and visualization tools like pajek, gephi, or VOS Viewer. However, many different algorithms for community extraction and topic detection exist and offer different suggestions what the most prominent groupings of publications or authors may be. The special session at ISSI 2015 sets out to systematically compare and evaluate the origin, extent, and implication of differences between topic extraction methods. In this paper we describe the results of our approach to topic detection and topic affinity analysis to the shared 'astronomy and astrophysics' data set. This approach has emerged from research program on studying behavioral patterns in scientific communities and comparing them across fields, and may help to shed light on the nature of differences between topic extraction solutions. Background As described in (Velden 2009), we take a mixed method approach to studying field-specific practices and cultures of scientific communities, integrating ethnographic field studies with network analytic methods. The network analytic method we apply here to the 'astronomy and astrophysics' data set is part of an ongoing effort to combine network analytic with ethnographic methods (Velden, Haque & Lagoze, 2010; Velden, 2013). This evolves a tradition of close-up analysis of scientific networks and communication practices started by Crane's work (1972) on invisible colleges and taken up more recently by Zuccala (2006).